Overfitting avoidance in genetic programming of polynomials Export

@inproceedings{citeulike:2615374, abstract = {This paper proposes several techniques for avoiding overfitting in the genetic programming (GP) of polynomials. The model specification flexibility is increased by: (1) a polynomial block reformulation, which reduces the statistical bias, and, (2) complexity tuning using local ridge regression and regularized weight subset selection, which reduce the statistical variance. Another contribution is the designed fitness function for search navigation towards highly predictive models. Experimental results on time-series forecasting show that these techniques help GP to find accurate, less complex and better forecasting polynomials than traditional Koza-style GP (J.R. Koza, 1992) and the previous Stroganoff system (H. Iba et al., 1994, 2001)}, author = {Nikolaev, N. and de Menezes, L. M. and Iba, H.}, booktitle = {Evolutionary Computation, 2002. CEC '02. Proceedings of the 2002 Congress on}, citeulike-article-id = {2615374}, citeulike-linkout-0 = {http://dx.doi.org/10.1109/CEC.2002.1004415}, citeulike-linkout-1 = {http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1004415}, doi = {10.1109/CEC.2002.1004415}, journal = {Evolutionary Computation, 2002. CEC '02. Proceedings of the 2002 Congress on}, keywords = {genetic-algorithm, genetic-programming, overfitting}, pages = {1209--1214}, posted-at = {2008-03-31 09:28:38}, priority = {2}, title = {Overfitting avoidance in genetic programming of polynomials}, url = {http://dx.doi.org/10.1109/CEC.2002.1004415}, volume = {2}, year = {2002} }

TY - CONF ID - citeulike:2615374 L3 - citeulike-article-id:2615374 N2 - This paper proposes several techniques for avoiding overfitting in the genetic programming (GP) of polynomials. The model specification flexibility is increased by: (1) a polynomial block reformulation, which reduces the statistical bias, and, (2) complexity tuning using local ridge regression and regularized weight subset selection, which reduce the statistical variance. Another contribution is the designed fitness function for search navigation towards highly predictive models. Experimental results on time-series forecasting show that these techniques help GP to find accurate, less complex and better forecasting polynomials than traditional Koza-style GP (J.R. Koza, 1992) and the previous Stroganoff system (H. Iba et al., 1994, 2001) JF - Evolutionary Computation, 2002. CEC '02. Proceedings of the 2002 Congress on EP - 1214 TI - Overfitting avoidance in genetic programming of polynomials VL - 2 T2 - Evolutionary Computation, 2002. CEC '02. Proceedings of the 2002 Congress on SP - 1209 KW - genetic-algorithm KW - genetic-programming KW - overfitting AU - Nikolaev, N AU - de Menezes, LM AU - Iba, H PY - 2002/// UR - http://dx.doi.org/10.1109/CEC.2002.1004415 ER -